支气管肺发育不良早产儿并发脑损伤的影响因素及预测模型

Influencing factors and prediction model of brain injury in preterm infants with bronchopulmonary dysplasia

摘要:

目的 探究早产儿支气管肺发育不良（BPD）并发脑损伤的影响因素并以此构建列线图预测模型并进行验证，以期为临床降低BPD早产儿脑损伤、改善预后提供参考。方法 回顾性分析2021年7月—2023年7月常州市妇幼保健院分娩的BPD早产儿98例，根据患儿是否出现脑损伤分为脑损伤组（30例）和无脑损伤组（68例）。分析影响患儿脑损伤的因素，并以此构建Nomogram列线图模型预测患儿脑损伤发生风险；绘制受试者工作特征（ROC）曲线，采用曲线下面积分析预测模型对患儿脑损伤的预测效能。结果 两组性别比例、纠正胎龄、分娩方式、胎儿数量、妊娠期高血压、妊娠期糖尿病、胎膜早破、羊水污染、败血症、听力损伤比较，差异均无统计学（P >0.05）。脑损伤组胎龄、出生体重低于无脑损伤组（P <0.05），出生5 min新生儿Apgar评分< 7分、机械通气时间≥ 7 d、代谢性酸中毒、新生儿感染占比均高于无脑损伤组（P <0.05）。多因素逐步Logistic回归分析结果显示：出生5 min新生儿Apgar评分［O^R=0.243（95% CI：0.119，0.494）］是患儿发生脑损伤的独立保护因素（P <0.05）；机械通气时间［O^R=3.567（95% CI：1.622，7.843）］、新生儿感染［O^R=5.339（95% CI：2.662，10.706）］是患儿发生脑损伤的独立危险因素（P <0.05）。基于上述影响因素构建的列线图预测模型经Bootstrap法内部验证显示C-index指数为0.832（95% CI：0.787，0.925），区分度良好。ROC曲线结果显示列线图模型预测患者SBI发生的敏感性为85.20%（95% CI：0.556，0.889）、特异性为88.70%（95% CI：0.620，0.915），曲线下面积为0.872（95% CI：0.787，0.957），模型预测效能良好。结论 出生5 min新生儿Apgar评分是BPD早产儿发生脑损伤的独立保护因素，机械通气时间、新生儿感染是独立危险因素，基于上述因素构建的列线图预测模型可以较好地评估患儿脑损伤发生风险。

Abstract:

Objective To explore the influencing factors of brain injury in preterm infants with bronchopulmonary dysplasia (BPD) and to construct and validate a nomogram prediction model, in order to provide guidance for reducing brain injury and improving prognosis in preterm infants with BPD.Methods Ninety-eight premature infants with BPD who were delivered from July 2021 to July 2023 in Changzhou Maternal and Child Health Hospital were retrospectively analyzed. According to whether the children had brain injury, they were divided into brain injury group (30 cases) and non-brain injury group (68 cases). The factors affecting brain injury were analyzed, based on which a nomogram model was constructed to predict the risk of brain injury in the infants. The area under receiver operating characteristic (ROC) curve (AUC) was used to assess the predictive efficacy of the model.Results There was no difference in the sex composition, corrected gestational age, modes of delivery, number of fetuses, and proportions of pregnancy-induced hypertension, gestational diabetes mellitus, premature rupture of membranes, amniotic fluid contamination, septicemia, and hearing impairment between the two groups (P > 0.05). The gestational age and the birth weight in the brain injury group were lower than those in the non-brain injury group (P < 0.05). The proportions of 5-minute Apgar score < 7, duration of mechanical ventilation ≥ 7 d, metabolic acidosis and neonatal infections in the brain injury group were all higher than those in the non-brain injury group (P < 0.05). Multivariable Logistic regression analysis showed that high 5-minute Apgar score [O^R = 0.243, (95% CI: 0.119, 0.494) ] was a protective factor for brain injury in neonates (P < 0.05). Long duration of mechanical ventilation [O^R = 3.567, (95% CI: 1.622, 7.843) ] and neonatal infection [O^R = 5.339, (95% CI: 2.662, 10.706) ] were risk factors for brain injury in neonates (P < 0.05). The verification of the nomogram prediction model constructed based on the above influencing factors by the Bootstrap method demonstrated that the C-index was 0.832 (95% CI: 0.787, 0.925), suggestive of great discrimination of the model. The ROC curve analysis revealed that the sensitivity of the nomogram prediction model for SBI was 85.20% (95% CI: 0.556, 0.889), with a specificity of 88.70% (95% CI: 0.620, 0.915) and an AUC of 0.872 (95% CI: 0.77, 0.957), indicating good predictive performance of the model.Conclusion High 5-minute Apgar score is an independent protective factor while long duration of mechanical ventilation and neonatal infection are independent risk factors for brain injury in preterm infants with BPD. The nomogram prediction model based on the above indicators can effectively assess the risk of brain injury in the neonates.